Evolution of biosilicification molecular toolkits in demosponges (Porifera)

Abstract Background Three out of the four classes of Porifera build siliceous skeletons, but they do this through divergent enzymatic pathways that are still poorly characterised. In demosponges, where the process is best known, the biosilicification occurs by polycondensation of silica using silicatein and other proteins (silica depolymerisers and scaffolding proteins) on an axis made of actin. For silicic acid intake, sponges use several types of membrane transporters with phylogenetic affinities to those in bacteria and plants. Most demosponges can produce more than one spicule type, sometimes beautifully ornamented, but this diversity is not clearly linked to the expression of polycondensing, scaffolding or transporter proteins. Here, we explore the evolution of the molecular toolkit for biosilicification in demosponges and across Porifera, using a time-calibrated phylogeny with an unparalleled dataset of 64 sponge genomes and transcriptomes. Results Our results indicate a significantly more complex toolkit of genes involved in the biosilicification process than previously described, with a wide array of clade-specific proteins, especially silicateins, which allows us to hypothesise a link between silicatein type and diversity of spicules. We also describe specific putative expansions of the molecular complements for water and silicic acid transporters (aquaporins), some being acquired through horizontal gene transfer, potentially linked to adaptations to freshwater conditions. Conclusions Our overview of the evolution of the biosilification process, with the most complete dataset to date, corroborates the hypothesis of independent origins for the main silicifying enzymes and provides a scenario of duplication and loss of genes that have allowed the massive diversification and evolutionary success of demosponges in a wide range of habitats.